Ernesto M. Morgado

Scheduling and managing crew in the Portuguese railways

Abstract We present a tool to generate applications in crew scheduling. The first application developed with this tool, called “ESCALAS”, schedules crews for the Portuguese Railways (it schedules both driving and other crew). This application is complete and installed in the Portuguese Railways. The application is presently undergoing extensive testing and has proved to produce schedules that are comparable with those produced by human experts, at a fraction of time. The system is planned to enter routine operation in May 1992. The system uses a graphic, highly intuitive interface, and enables four different modes of operation: (1) Manual operation, in which the user makes the decisions in building the schedule and the system automatically verifies all the constraints and labour rules imposed upon it; (2) Semiautomatic operation, in which the system gives hints about hoe the user should pursue the schedule; (3) Fully automatic operation, in which thesystem automatically builds a schedule, according to a predefined set of scheduling strategies; (4) Mixed mode, in which there is an interaction with the user who may impose certain restrictions or combinations of tasks, the systems partially builds a schedule which can then be modified by the user and the automatic generation may proceed with the restrictions imposed. Besides, the system is able to perform crew assignment to each one of the schedules produced and can modify both the schedule and the assignment produced.

Optimising Cyclic Timetables with a SAT Approach

This paper describes the preliminary results of an ongoing research on cyclic railway timetabling, namely on optimising timetables with respect to travel time using Boolean Satisfiability Problem (SAT) approaches.

Managing Operations In Real-time

This paper describes a real-time dispatching system that allows real-time monitoring of train operations. The system was developed by SISCOG. It also offers optimizers for use in dealing with disruptions in train operations.

Solving Set Partitioning Problems with Global Constraint Propagation

We propose a constraint-based approach for solving set partitioning problems. We show that an efficient, open and easily modifying model is obtained by using a constraint propagator that is: global, in the sense that it enforces consistency between local knowledge (such as variable domains) and global knowledge (such as the optimisation goal); and dynamic, in the sense that it propagates the decisions taken during the search process. This propagator derives new constraints based on the existing ones by efficiently chaining a set of propagation rules that we present here and demonstrate. This propagator can be used not only to prune efficiently the search space, but also to prove in certain cases that a given solution is optimal. This approach was tested with five crew duty scheduling problems supplied by two operators from the railway and bus domains. Results were compared with the ones obtained with an approach that is a good representative of the industrial state-of-the-art.